A frequency choice for the SPL machine: Impact on hardware O. Brunner/ E. Ciapala

General considerations: Nb. of SC cavities: 234 Power per cavity: 1MW (high β region) RF pulse length: 1.55ms / 50Hz - 40mA (HPSPL) 1.90ms / 2Hz - 20mA (LPSPL) Remark: - low β region: less power per cavity required  more cavities per klystron - HPSPL vs LPSPL: 1MW, 1.5ms pulse, 50Hz instead of 500kW , 1.9ms, 2Hz per cavity  much higher average power + power distribution scheme affected

Klystrons: 704MHz 805MHz SNS klystron (TH2168) Ucat/ Icat : 140kV/88A duty cycle: 1.5ms / 60Hz gain / η: 50dB / 50% av. Power 450kW Cost estimate: 290kEuros/tube Modulator: SNS HVCM 10MW 704MHz mutli-beam klystron 1.4GHz 1.3GHz ILC – FNAL-KEK (TH2104) Ucat/ Icat : 128kV/88A duty cycle: 2ms / 10Hz gain / η: 50dB / 45% av. Power 100kW Cost estimate: 240kEuros/tube Modulator: similar to SNS 10MW 1.3GHz MBK – FLASH/XFEL Ucat/ Icat : 140kV/155A duty cycle: 1.5ms / 10Hz gain / η: 50dB / 50% av. Power 150kW Price estimation: 400kEuros/tube Modulator: new development at 10Hz for XFEL 5MW option: 4 (or 8) cavities per klystron Pav ~450kW feasible??? 10MW option: 8 (or 16) cavities per klystron Pav ~450kW feasible??? FEASIBLE??? Development cost for new design 1MEuros/vendor and takes 1-2 years (applies to all options)

Klystrons – some considerations: Electron gun design: cathode-to-beam convergence factor increases with frequencies not a problem with MBKs (lower gun perveance) Beam perveance: klystrons efficiency increases with lower perveance i.e.: lower charge density = easier bunching  better energy transfer from DC to AC. Output window: peak power limitation – doesn’t seem to be a issue in our case. MBKs have two windows. Collector: most effective design take advantage of the latent heat of water evaporation by allowing boiling at the collector surface (hypervapotron), can absorb: ~3kW/cm2  maximizing the wetted area  optimizing the spread of the focused beam on the collector surface Higher power levels @ lower frequencies Advantage of the MBK Higher power levels @ lower frequencies

For the modulators power is important, not frequency 5MW klystrons SNS HVCM (Los Alamos) Rectangular HV pulses 140kV @ 90A, 60Hz 11MW peak power, 1MW average 8% max duty cycle trise – tfall: 100µs ripple ~0.1% modularity for different klystron configuration 15 HVCM installed Cost estimate: ? More info on real tests needed 10MW klystrons Will this work @ 50Hz? (XFEL will push the rep rate to 30Hz – but reduce pulse length to keep av. power < 250kW) Bouncer modulator for XFEL & FLASH Rectangular HV pulses 120kV @ 140A, 10Hz 3 systems supplied by FNAL for TESLA (now in operation at FLASH) 8 systems made in cooperation with industry. New development by industry for XFEL: bouncer type + pulse step – to be tested in 2008 Cost estimate:? For the modulators power is important, not frequency

RF distribution: Examples with 5MW klystrons feeding 8 cavities (LPSPL) or 4 cavities (HPSPL) (10MW klystrons would feed respectively 16 or 8 cavities) XFEL linear distribution Asymmetric shunt tees (can be pretuned with coupling ratio 1-9dB) TTF/FLASH/ILC linear distribution Tree like distribution with circulators Tree like distribution without circulators Shopping list: WR1150 or WR650 1 circulator, RF loads per cavity 1vector modulator or 1 phase shifter + Qext transformer per cavity Big 5MW circulators & loads if tree like distribution with big protecting circulators and magic T’s envisaged (MBKs: 2 output windows)

RF components: Power requirements: 704MHz 1.4GHz LPSPL: 0.5MW/cavity, 1.9ms RF pulse, 2Hz  Pav ~2kW HPSPL: 1MW/cavity, 1.5ms RF pulse, 50Hz  Pav ~75kW 704MHz WG losses lower for the larger WR 1150 WG (or WR1500??) SF6 5MW circulators & RF (water) loads exist (SNS) 1MW circulators & RF ferrite loads exist (APT) 1 MW vector modulator easier to built @ 704MHz? 75kW average power ferrite load – looks feasible length ~1500mm 1.4GHz Smaller WG components cost less SF6 New and sophisticated WR650 WG components developed for TTF, ILC, XFEL (e.g. circulators, loads, tuners, couplers,…) however only @ ~280kW, 1.5ms,10Hz  ~ 4-5kW average RF loads: A ferrite load would be about 3 meters long (width 14cm)!! ferrite thickness/material in the first section is an issue Water load: Existing loads ? (Cornell: 20kW CW loads 1.3GHz) Existing circulators @ 1.3GHz ILC type: problems >600us (AFT) 20kW CW (Cornell) 5MW, 50kW average (huge/expensive device) Not realistic - WG components do NOT all exist yet - Different RF distribution scheme can be envisaged

RF power couplers: Power requirements: 704MHz 1.4GHz LPSPL: 0.5MW/cavity, 1.9ms RF pulse, 2Hz  Pav ~2kW HPSPL: 1MW/cavity, 1.5ms RF pulse, 50Hz  Pav ~75kW 704MHz HIPPI 1MW power coupler 1MW peak, 10% duty cycle KEK/SNS type: Coaxial warm window He cooled coaxial lower part No bellows Qext = 106 1.4GHz Examples of existing couplers: XFEL power coupler: ~280kW, 10Hz ERL: 1.5GHz ~35kW CW (tested up 50kW) From Sergey Belomestnykh we learned that "their modified TTF coupler is capable of handling up to 75 kW CW (they were tested up to 61 kW and should support up to 75). Since no major design change was done, the original peak power of 1.5 MW should still be valid“ (SPL Steering group minutes 7.12.08) …does not seem to be a unanimous agreement tests to be completed soon more info and/or R&D needed

Some personal conclusions: 704MHz Pros: Potentially higher klystrons peak &average power levels Reduced number of power stations Average power availability of passive components More robust, less sensitive and less challenging Cons: Reduced size of waveguides, cavities Availability of components on the market Expertise around the world 1.4GHz Pros: Reduced size of waveguides, cavities Availability of components on the market Expertise around the world Cons: Limited average power level &Extreme high power density Trade off between peak power levels and duty cycle Same arguments –to a lower extend- apply for LPSPL